I have spent more than twenty years working in technology. Technical knowledge has helped me build a business, solve problems, and protect organizations. I understand its value because I have built much of my professional life around it.
That is also why I believe we need to reconsider what we tell people about preparing for the future.
For years, the advice sounded straightforward. Study science, technology, engineering, or mathematics. Learn to code. Develop a technical skill. Choose a degree with an obvious connection to a job.
Meanwhile, students interested in history, philosophy, literature, or other humanities fields often faced a different question: “What are you going to do with that?”
Artificial intelligence is giving us a reason to revisit that conversation.
A recent Forbes article by Lisa Chambers explores the possibility of a humanities comeback as AI changes the skills employers value. A broad recovery remains uncertain, but the premise deserves attention: the ability to interpret, question, communicate, and understand people may become increasingly valuable as technology handles more of the production.
My view is that the biggest change reaches well beyond college majors.
When more people can produce an answer, the advantage shifts toward those who can determine whether it is the right answer to the right question.
The Economics of Getting Things Done Are Changing
Much of professional work has traditionally required both knowing what needs to happen and possessing the skills to make it happen.
You might understand what you want a software application to accomplish but lack the ability to build it. You might have a business idea but struggle to turn it into a proposal. You might recognize a problem without knowing how to organize the information needed to investigate it.
AI can shorten the distance between intention and a usable first attempt.
It can help draft the proposal, generate code, summarize information, explore alternatives, or turn rough notes into a plan. The quality varies, and the results require review. Still, these capabilities change who can participate in work that previously demanded substantial specialized training.
That does not make expertise irrelevant. It changes where expertise contributes the most.
When producing a first draft takes less time, understanding the assignment becomes a larger part of the value. When creating several options becomes easier, choosing among them matters more.
The difficult questions remain: What problem are we solving? What assumptions have we made? What does success actually mean? Who will be affected if we get this wrong?
A tool can help us examine those questions. Responsibility for the answers stays with us.
Why the Humanities Belong in This Conversation
At their best, the humanities require people to work through ambiguity.
History asks us to evaluate sources, examine context, and consider competing explanations. Philosophy requires us to define our terms, test an argument, and confront contradictions. Literature asks us to interpret language, recognize perspective, and understand motivations that may be very different from our own.
Those habits have practical applications.
A leader evaluating an AI-generated recommendation needs to recognize an unsupported assumption. A manager introducing a new process needs to understand why employees might resist it. A business owner reviewing a persuasive analysis needs to distinguish evidence from confident language.
None of these abilities belongs exclusively to a particular degree. Engineers can be excellent critical thinkers and communicators. Humanities graduates can be careless, incurious, or unwilling to challenge their beliefs.
The credential does not guarantee the capability.
But we should recognize the capability when it exists. Careful reading, clear writing, ethical reasoning, and an understanding of human behavior deserve a place beside technical competence.
Calling them “soft skills” can make them sound optional. Anyone who has tried to lead people through a difficult change knows how demanding they can be.
Clear Language Becomes Part of Execution
Working with AI makes something visible that has always mattered in leadership: unclear thinking produces unclear instructions.
Consider a manager who asks AI to “improve customer service.”
Improve it how? Reduce response times? Resolve more issues during the first interaction? Help customers understand their options? Reduce the number of times they have to repeat themselves?
Each interpretation leads to different recommendations.
Useful direction requires context, priorities, constraints, and a definition of success. Someone has to explain what matters and why.
That makes writing more than a way to communicate a finished thought. Writing becomes a way to discover whether the thought is finished.
When you try to explain an idea precisely, gaps appear. You realize that you have not defined the problem, resolved a conflict between priorities, or identified the evidence needed to support your position.
AI can help with that process. It can ask questions, identify inconsistencies, and suggest other interpretations. But the person using it must care enough to work through those issues.
A polished response does not fix a poorly understood problem.
AI Can Expand Our Thinking or Help Us Avoid It
This is the part of the conversation that concerns me most.
AI gives us an opportunity to examine more perspectives, test more ideas, and challenge our assumptions. It also gives us an easy way to skip that work.
We can ask it to argue against our position. We can ask what evidence would change the recommendation. We can explore the consequences of a decision for people whose interests differ from our own.
Or we can ask it to make our existing opinion sound intelligent.
Both uses are convenient. Only one meaningfully improves our judgment.
There is a real temptation to confuse the experience of receiving a coherent answer with the experience of understanding something. We read a confident explanation, recognize familiar language, and move on.
But could we explain the reasoning ourselves? Could we identify its weakest assumption? Would we notice if an important fact were missing?
AI can support the work of thinking. It can also make the absence of thinking harder to notice.
That distinction will matter in classrooms, boardrooms, and everyday decisions.
Faster Execution Raises the Cost of Poor Judgment
Imagine a company using AI to improve its customer support process.
The system helps draft procedures, categorize requests, and recommend ways to reduce the time spent on each interaction. Everything looks organized. The language is professional. The proposed workflow appears efficient.
But suppose leadership defined success entirely around closing tickets faster.
The company could become better at ending interactions while customers become more frustrated with the results.
Someone has to recognize that speed is only one part of service. Customers also need resolution, clarity, confidence, and an understanding of what happens next.
That requires judgment about the purpose of the work.
In technology, it is easy to become focused on whether something can be implemented. Leadership also requires us to consider whether it should be implemented, under what conditions, and with what consequences.
AI can accelerate an effective process. It can also accelerate a process built around the wrong objective.
The faster we can act, the more deliberate we need to be about direction.
Technical Depth Still Matters
I would be cautious about interpreting this shift as a reason to abandon STEM.
Someone still needs to understand the systems, validate the calculations, test the software, and identify failures that a convincing explanation might conceal.
AI-generated code does not remove the need for someone who understands security, architecture, and maintenance. An attractive analysis still needs someone who can assess the data and methods behind it.
Humanities work faces its own pressures. AI can generate prose and summarize texts, so a degree centered on writing offers no automatic protection from disruption.
The strongest preparation, in my view, combines depth in a field with the ability to work across it.
A technical professional who understands people can connect a solution to a real need. A humanities graduate with technical fluency can translate insight into something useful and assess the tools involved.
Both need practice taking an idea through to a result.
Good intentions and thoughtful discussion are not enough. Neither is rapid production without understanding.
Education Has to Preserve the Work of Learning
This creates a challenge for education.
If AI can complete an assignment, schools need to be clear about what the assignment is supposed to develop.
Writing an essay has value beyond the finished document. The student has to organize evidence, make choices, build an argument, and revise weak reasoning. Solving a technical problem develops understanding through the struggle to reach an answer.
If we automate away all of that struggle, we may produce better-looking assignments while students learn less.
Students need opportunities to use AI. They also need to explain their reasoning, defend their conclusions, inspect sources, and demonstrate that they understand the work carrying their name.
The same principle applies inside a business.
A teammate who can generate a report has completed one part of the job. A teammate who can explain its limitations, connect it to the business, and recommend a defensible next step has contributed something much more valuable.
We Still Own the Outcome
The prospect of a humanities comeback interests me because it challenges a narrow definition of useful knowledge.
We need people who can build and maintain technology. We also need people who can understand its consequences, question its assumptions, and help others use it well.
Often, we need those abilities in the same person.
AI gives us more capacity to create, analyze, and act. What we do with that capacity depends on the quality of our thinking and the standards we bring to the work.
A customer still needs help. A teammate still needs direction. A decision still affects real people.
The tool may do more of the execution.
We still have to understand the problem, choose the direction, and own the result.